target: fix a race condition between login_work and the login thread
PATCH v2 1/1 target: fix a race condition between login_work and the login thread
In case a malicious initiator sends some random data immediately after a login PDU; the iscsi_target_sk_data_ready() callback will schedule the login_work and, at the same time,
the negotiation may end without clearing the LOGIN_FLAGS_INITIAL_PDU flag (because no additional PDU exchanges are required to complete the login).
The login has been completed but the login_work function will find the LOGIN_FLAGS_INITIAL_PDU flag set and will never stop from rescheduling itself;
at this point, if the initiator drops the connection, the iscsit_conn structure will be freed, login_work will dereference a released socket structure and the kernel crashes.
BUG: kernel NULL pointer dereference, address: 0000000000000230 PF: supervisor write access in kernel mode PF: error_code(0x0002) - not-present page Workqueue: events iscsi_target_do_login_rx [iscsi_target_mod] RIP: 0010:_raw_read_lock_bh+0x15/0x30 Call trace: iscsi_target_do_login_rx+0x75/0x3f0 [iscsi_target_mod] process_one_work+0x1e8/0x3c0
Fix this bug by forcing login_work to stop after the login has been completed and the socket callbacks have been restored. Also fix other potential race conditions in the error paths.
V2: remove an unnecessary call to cancel_delayed_work();
fix a potential race condition in iscsi_start_negotiation() and in iscsi_target_do_login_rx()'s error paths
--- drivers/target/iscsi/iscsi_target_nego.c | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-)
drivers/target/iscsi/iscsi_target_nego.c +3 -3
@@ -645,7 +645,6 @@ static void iscsi_target_do_login_rx(struct work_struct *work) LOGIN_FLAGS_WRITE_ACTIVE)) goto err; }else if (rc ==1 ) { -cancel_delayed_work (&conn->login_work);iscsi_target_nego_release (conn);iscsi_post_login_handler (np, conn, zero_tsih);iscsit_deaccess_np (np, tpg, tpg_np);
@@ -654,7 +653,7 @@ static void iscsi_target_do_login_rx(struct work_struct *work) err:iscsi_target_restore_sock_callbacks (conn); -cancel_delayed_work (&conn->login_work); +cancel_delayed_work_sync (&conn->login_work);iscsi_target_login_drop (conn, login);iscsit_deaccess_np (np, tpg, tpg_np); }
@@ -1058,6 +1057,7 @@ static int iscsi_target_do_login(struct iscsit_conn *conn, struct iscsi_login *l login->tsih = conn->sess->tsih; login->login_complete =1 ;iscsi_target_restore_sock_callbacks (conn); +cancel_delayed_work_sync (&conn->login_work);if (iscsi_target_do_tx_login_io (conn, login) <0 )return -1 ;
@@ -1363,8 +1363,8 @@ int iscsi_target_start_negotiation( ret =-1 ;if (ret <0 ) { -cancel_delayed_work_sync (&conn->login_work);iscsi_target_restore_sock_callbacks (conn); +cancel_delayed_work_sync (&conn->login_work);iscsi_remove_failed_auth_entry (conn); }if (ret !=0 )
Signature · 1 line
2.31.1
pá 11. 11. 2022 v 16:47 odesílatel Maurizio Lombardi <[email protected]> napsal:
Quoted 5 lines
Fix this bug by forcing
login_workto stop after the login has been completed and the socket callbacks have been restored. Also fix other potential race conditions in the error paths.
Self-NACK
I can't call cancel_delayed_work_sync() in iscsi_target_do_login() because the latter could be called from the login_work context and this will cause a deadlock.
Will submit a V3 when ready
Quoted 52 lines
V2: remove an unnecessary call to
cancel_delayed_work();
fix a potential race condition iniscsi_start_negotiation()and in iscsi_target_do_login_rx()'s error pathsSigned-off-by Maurizio Lombardi <[email protected]>--- drivers/target/iscsi/iscsi_target_nego.c | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-)drivers/target/iscsi/iscsi_target_nego.c +3
-4@@ -645,7 +645,6 @@ static void iscsi_target_do_login_rx(struct work_struct *work) LOGIN_FLAGS_WRITE_ACTIVE)) goto err; }else if (rc ==1 ) { -cancel_delayed_work (&conn->login_work);iscsi_target_nego_release (conn);iscsi_post_login_handler (np, conn, zero_tsih);iscsit_deaccess_np (np, tpg, tpg_np);@@ -654,7 +653,7 @@ static void iscsi_target_do_login_rx(struct work_struct *work) err:iscsi_target_restore_sock_callbacks (conn); -cancel_delayed_work (&conn->login_work); +cancel_delayed_work_sync (&conn->login_work);iscsi_target_login_drop (conn, login);iscsit_deaccess_np (np, tpg, tpg_np); }@@ -1058,6 +1057,7 @@ static int iscsi_target_do_login(struct iscsit_conn *conn, struct iscsi_login *l login->tsih = conn->sess->tsih; login->login_complete =1 ;iscsi_target_restore_sock_callbacks (conn); +cancel_delayed_work_sync (&conn->login_work);if (iscsi_target_do_tx_login_io (conn, login) <0 )return -1 ;@@ -1363,8 +1363,8 @@ int iscsi_target_start_negotiation( ret =-1 ;if (ret <0 ) { -cancel_delayed_work_sync (&conn->login_work);iscsi_target_restore_sock_callbacks (conn); +cancel_delayed_work_sync (&conn->login_work);iscsi_remove_failed_auth_entry (conn); }if (ret !=0 ) --2.31.1
út 15. 11. 2022 v 9:33 odesílatel Maurizio Lombardi <[email protected]> napsal:
Quoted 9 lines
pá 11. 11. 2022 v 16:47 odesílatel Maurizio Lombardi <[email protected]> napsal:
Quoted 5 lines
Fix this bug by forcing
login_workto stop after the login has been completed and the socket callbacks have been restored. Also fix other potential race conditions in the error paths.
Quoted 7 lines
Self-NACK
I can't call
cancel_delayed_work_sync()iniscsi_target_do_login()because the latter could be called from thelogin_workcontext and this will cause a deadlock.Will submit a V3 when ready
The correct place where to call cancel_delayed_work_sync() is in iscsi_target_start_negotiation(), because the latter is only called in the login_thread context.
I'm going to test the following:
(quoted diff) +0 -0
b/drivers/target/iscsi/iscsi_target_nego.c
index f2919319ad38..b9bd77f41c8d 100644
drivers/target/iscsi/iscsi_target_nego.c +4 -2
@@ -1363,12 +1363,14 @@ int iscsi_target_start_negotiation( ret =-1 ;if (ret <0 ) { -cancel_delayed_work_sync (&conn->login_work);iscsi_target_restore_sock_callbacks (conn); +cancel_delayed_work_sync (&conn->login_work);iscsi_remove_failed_auth_entry (conn); } -if (ret !=0 ) +if (ret !=0 ) { +cancel_delayed_work_sync (&conn->login_work);iscsi_target_nego_release (conn); + }return ret; }
Maurizio